Task-oriented cooperative low-energy consumption fast routing method for unmanned aerial vehicle

Abstract

The application discloses a task coordination-oriented low-energy-consumption fast routing method for unmanned aerial vehicles, and belongs to the fields of wireless communication, unmanned aerial vehicle ad hoc networks and routing protocols. In view of the unmanned aerial vehicle routing problem under different network topological environments and different moving speeds, the application provides a task coordination-oriented low-energy-consumption fast routing method for unmanned aerial vehicles. The application enables the unmanned aerial vehicles to realize distributed routing, utilizes the channel state of the unmanned aerial vehicle group and the routing experience parameters of adjacent unmanned aerial vehicles and other information, adopts a reinforcement learning algorithm, optimizes the routing decision and transmission power of the unmanned aerial vehicles, satisfies the end-to-end delay constraint, and can carry out search and rescue and target tracking and other tasks based on sensing data and control messages. The application effectively improves the routing stability of the high-speed moving unmanned aerial vehicle group, reduces the end-to-end delay and reduces the routing energy consumption.

Description

Technical Field

[0001] This invention belongs to the fields of wireless communication, UAV ad hoc networks and routing protocols, specifically relating to a low-power and fast routing method for UAVs oriented towards task collaboration. Background Technology

[0002] Unmanned aerial vehicle (UAV) ad hoc networks (UAVs) feature fast-moving nodes with limited energy and highly variable network topologies. Upper-layer applications such as UAV control, search and rescue, and target tracking place higher demands on transmission reliability and latency. However, existing ad hoc network routing protocols rely on route discovery processes or location information. Under rapidly fluctuating channel conditions, routing information estimated based on average channel quality information may become outdated, while location-based routing protocols struggle to obtain distances to destinations in large-scale dynamic networks. In scenarios with significant node movement, flooding across the entire network is the most effective routing method. However, simple flooding often leads to congestion and conflicts due to excessive redundant broadcasts, resulting in broadcast storms. Therefore, a multi-path routing method that eliminates the need to maintain a global routing table and avoids numerous invalid broadcasts is crucial for ensuring data transmission in UAV ad hoc networks.

[0003] UAV ad hoc network routing improves upon traditional routing protocols, reducing the cost of maintaining the topology while enhancing flexibility in responding to dynamic environments. S. Rosati et al. [S. Rosati, K. Kru˙zelecki, G. Heitz, et al., Dynamic routing for flying Ad-hoc networks, IEEE Transactions on Vehicular Technology, vol. 65, no. 3, pp. 1690-1700, Mar. 2016] proposed a predictive optimization link state routing method. This method predicts the communication quality of the wireless link based on the UAV's GPS location information, adaptively adjusts the expected transmission count metric, and tracks topology changes, significantly reducing the probability of outages compared to the traditional OLSR protocol. Chinese patent application publication number CN115087065A proposes a location prediction-based UAV formation routing protocol. This protocol calculates the maximum communication distance between UAVs, predicts their movement trajectories and calculates the communication probability, comprehensively evaluates multiple alternative paths, and improves the success rate of data transmission.

[0004] Selective broadcast routing is a multipath routing method that allows data packets to be forwarded to their destination via multiple shortest paths, reducing network latency and energy consumption. H. Song et al. [SH Song, L. Liu, B. Shang, et al., Enhanced flooding-based routing protocol for swarm UAV networks: random network coding met clustering, Proc. IEEE Int. Conf. Comput. Commun. (INFOCOM), Vancouver, BC, Canada, May 2021, pp. 1–10] proposed a flooding routing protocol based on random coding and clustering, significantly accelerating the flooding-based routing process. Furthermore, it broadcasts only through cluster heads, effectively preventing broadcast storms and ACK feedback conflicts. Chinese patent application publication number CN111542097A proposes a Q-learning-based broadcast routing algorithm that determines whether to broadcast based on the proportion of nodes that have already received data among all of a node's neighbors, avoiding excessive redundant broadcasts and saving energy.

[0005] Reinforcement learning-based adaptive routing schemes can better adapt to the complex environment of UAV ad hoc networks. Z. Zheng et al. [Z. Zheng, AK Sangaiah, and T. Wang, Adaptive communication protocols in flying Ad-hoc network, IEEE Communications Magazine, vol. 56, no. 1, pp. 136–142, Jan. 2018] utilize UAV location information and a reward function defined based on global network utility to update local routing policies, reducing packet transmission latency. Chinese patent application publication number CN114449608A proposes a Q-learning-based adaptive routing method for UAV ad hoc networks. In the route discovery phase, the Q-table is updated based on link quality to establish routing paths. In the route maintenance phase, HELLO messages are used to detect network topology changes, and the Q-table is dynamically updated to maintain routing paths, reducing routing latency. Chinese patent application publication number CN112822752A proposes a routing establishment method and system for unmanned aerial vehicle (UAV) self-organizing networks. The method dynamically selects the optimal clustering strategy based on feedback information to obtain a stable cluster structure that adapts to the dynamic changes in network status, thereby extending the network's lifespan. Summary of the Invention

[0006] The purpose of this invention is to provide a low-energy, fast UAV routing method oriented towards task collaboration, addressing the routing problem of UAVs in different network topologies and with varying movement speeds. It utilizes reinforcement learning algorithms to dynamically optimize routing decisions and transmission power in UAV ad hoc networks, adapting to the frequent communication link breaks characteristic of highly dynamic UAV ad hoc networks. This effectively improves the routing stability of high-speed mobile UAV swarms, reduces end-to-end latency, and minimizes routing energy consumption.

[0007] This invention includes the following steps:

[0008] Step 1: In the drone self-organizing network The number of data packets that need to be transmitted by the drone is: The time it takes to transmit one data packet is called one time slot, and the total number of time slots is [number missing]. Routing decisions based on CSMA / CA contention mechanism ,when Do not broadcast data packets, set up drones The number of supported transmission powers is ,when Transmission power Then the optional action ,in, Indicates the maximum transmission power;

[0009] Step 2: Initialize the maximum routing efficiency weight Routing efficiency weight update rate The future benefits of the route learning process The weight of long-term returns shared by neighbors Delay risk level Risk threshold Latency risk learning rate State dimension This includes the maximum dimension of channel gain. , indivual Value matrix and Value matrix The weight of risk value in routing strategy selection The drone's own battery power Channel gain The set of received data packets , number of jumps Movement factor The number of times the monitored data packets were rebroadcast and initial performance signal-to-noise ratio and end-to-end delay ;

[0010] Step 3: In the Time slots, drones Received drone Broadcast data packets Determine the tuple Does it belong to the set of received data packets? ,in For the source drone ID, This is the data packet sequence number, which increments with each data packet sent by the source drone; if the tuple Belongs to the data packet set If the packet is not broadcast repeatedly, discard it to avoid repeated broadcasts; otherwise, proceed with the following routing decision steps.

[0011] Step 4: In the Time slots, drones Estimate the relationship with neighboring drones Channel gain Assess your remaining battery power. Signal-to-noise ratio of the received signal and data packet size Get the number of hops the current data packet has traversed from the IP packet header. Listen for the number of rebroadcasts of the previous data packet among the neighbors. Before obtaining destination feedback Average end-to-end latency of data packets Obtain the number of neighbors by interacting with neighbor beacons. Calculation of drones Compared to drones common neighbors set , Represent the neighbor set and calculate the mobility factor compared to the previous time slot. ;

[0012] Step 5: Calculate the single-hop delay including contention and transmission delay. Channel gain Signal-to-noise ratio Single-hop delay and the number of rebroadcasts detected Encapsulated in the MAC packet header, the data packet is obtained through neighbor replay. Received signal signal-to-noise ratio Single-hop delay Statistics and Neighbors Channel gain Accumulate and update the number of rebroadcasts ;

[0013] Step 6: Build the routing state ;

[0014] Step 7: Configure routing status Input benefit table and risk table Obtain the long-term reward value of the state-action pair. and risk value Update strategy distribution :

[0015]

[0016] Step 8: Drone According to strategy distribution Perform routing decisions and power allocation, then process the tuples. Store in collection ;

[0017] Step 9: After the target drone receives the data packet, it performs statistics. Average end-to-end latency of data packets The performance evaluation metrics are then sent to each drone via a reliable feedback channel; the drones Upon receiving routing performance metrics, it estimates its own transmission energy consumption. Calculate current benefits :

[0018]

[0019] in, This is a packet arrival indicator; if the data packet successfully reaches the destination drone, then... Otherwise, it is 0; and These represent the weighting factors for latency and energy consumption, respectively.

[0020] Step 10: Drone Receive neighbor set Shared state-value function, updating the Q-table:

[0021]

[0022] The dynamic routing efficiency weight According to the movement factor Perform the calculation:

[0023]

[0024] Step 11: Drone Divide end-to-end delay into Risk Level , and the previous The latency of each data packet is compared, and the proportion of packets with latency less than the risk threshold is counted to obtain the latency tolerance probability. Based on the minimum latency tolerance... Calculate latency risk :

[0025]

[0026] in, This indicates an indicator function; it returns 1 if the condition in parentheses is true, and 0 otherwise; based on latency risk... Update table E:

[0027]

[0028] Step 12: Repeat steps 3 to 11 until the algorithm converges.

[0029] Compared with the prior art, the present invention has the following outstanding advantages:

[0030] This invention enables UAVs to continuously optimize routing strategies based on factors such as packet size, channel gain with neighbors, and the number of replays of monitored packets. Considering risk values ​​based on latency constraints and experience sharing based on distributed value functions, a modified Boltzmann distribution is designed. In latency-sensitive applications, this reduces the exploration of high-latency risk strategies. It utilizes learned parameters shared among neighbors to accelerate the routing process and improve path stability, effectively reducing end-to-end latency and routing energy consumption. This provides fast and efficient communication and data transmission support for upper-layer applications such as UAV control, search and rescue, and target tracking. This invention empowers UAVs to achieve distributed routing. Utilizing information such as the channel state of the UAV swarm and routing experience parameters of neighboring UAVs, it employs reinforcement learning algorithms to optimize UAV routing decisions and transmission power, satisfying end-to-end latency constraints. It can support tasks such as search and rescue and target tracking based on sensing data and control messages. This invention effectively improves the routing stability of high-speed mobile UAV swarms, reduces end-to-end latency, and decreases routing energy consumption. Attached Figure Description

[0031] Figure 1 This refers to the end-to-end delay of the routing method described in this embodiment of the invention.

[0032] Figure 2 This refers to the routing energy consumption of the routing method described in the embodiments of the present invention. Detailed Implementation

[0033] To better understand the technical content of this invention, the technical solution of this invention will be described below in conjunction with the following specific implementation examples.

[0034] The embodiments of the present invention include the following steps:

[0035] Step 1: Set up 20 drones, requiring the transmission of 1200 data packets. The transmission time for one data packet is one time slot, resulting in a total of 1200 time slots. Routing decisions are made based on the CSMA / CA contention mechanism. ,when The transmit power available for relay drones can be selected without broadcasting data packets. The number is 4, let the th... Transmit power of time slot Then the optional action ,in .

[0036] Step 2: Set the weight for maximizing routing efficiency Routing efficiency weight update rate The future benefits of the route learning process The weight of long-term returns shared by neighbors Delay risk level Risk threshold Latency risk learning rate , indivual Value matrix and Value matrix The weight of risk value in routing strategy selection The drone's own battery power Channel gain The set of received data packets , number of jumps Movement factor The number of times the packet rebroadcast was detected and initial performance signal-to-noise ratio and end-to-end delay .

[0037] Step 3: In the Time slots, drones Received drone Broadcast data packets Determine tuples Does it belong to the set of received data packets? ,in For the source drone ID, This is the data packet sequence number, which increments with each data packet sent by the source drone. If the tuple... Belongs to the data packet set If the packet is not broadcast repeatedly, discard it to avoid repeated broadcasts; otherwise, proceed with the following routing decision steps.

[0038] Step 4: In the Time slots, drones Estimate the relationship with neighboring drones Channel gain Assess your remaining battery power. Signal-to-noise ratio of the received signal and data packet size Get the number of hops the current data packet has traversed from the IP packet header. Listen for the number of rebroadcasts of the previous data packet among the neighbors. Before obtaining destination feedback Average end-to-end latency of data packets Obtain the number of neighbors by interacting with neighbor beacons. Calculation of drones Compared to drones common neighbors set , Represent the neighbor set and calculate the mobility factor compared to the previous time slot. .

[0039] Step 5: Calculate the single-hop delay including contention and transmission delay. Channel gain Signal-to-noise ratio Single-hop delay and the number of rebroadcasts detected Encapsulated in the MAC packet header, the data packet is obtained through neighbor replay. Received signal signal-to-noise ratio Single-hop delay Statistics and Neighbors Channel gain Quantize it into 5 orders, accumulate and update the number of rebroadcasts. .

[0040] Step 6: Build the routing state .

[0041] Step 7: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full Input benefit table and risk table Obtain the long-term reward value of the state-action pair. and risk value Update strategy distribution :

[0042]

[0043] Step 8: Drone According to strategy distribution Perform routing decisions and power allocation, then process the tuples. Store in collection .

[0044] Step 9: After the target drone receives the data packet, it performs statistics. Average end-to-end latency of data packets The performance evaluation metrics are then sent to each drone via a reliable feedback channel. Upon receiving routing performance metrics, it estimates its own transmission energy consumption. Calculate current benefits :

[0045]

[0046] in This is a packet arrival indicator; if the data packet successfully reaches the destination drone, then... Otherwise, it is 0. Let the delay weighting factor be... Energy consumption weighting factor .

[0047] Step 10: Drone Receive neighbor set Shared state-value function, updating the Q-table:

[0048]

[0049] Among them, dynamic routing benefit weight According to the movement factor Perform the calculation:

[0050]

[0051] Step 11: Drone Divide end-to-end delay into Risk Level , and the previous The latency of each data packet is compared, and the proportion of packets with latency less than the risk threshold is counted to obtain the latency tolerance probability. Then, the minimum latency tolerance is determined. Calculate latency risk :

[0052]

[0053] in, This indicates an indicator function; it returns 1 if the condition in parentheses is true, and 0 otherwise. (Based on latency risk) Update table E:

[0054]

[0055] Step 12: Repeat steps 3 to 11 until the algorithm converges.

[0056] Depend on Figures 1-2 As can be seen, the embodiments of the present invention can reduce the end-to-end latency and routing energy consumption of drones.

[0057] This invention proposes a low-energy, fast routing method for UAVs oriented towards task collaboration. It estimates the channel state between UAVs and neighboring UAVs, assesses the UAV's energy and the number of neighboring UAVs, obtains the current route hop count and packet replay count, and derives the average end-to-end latency from feedback information from the destination UAV. A secure reinforcement learning algorithm is employed to dynamically optimize UAV node routing decisions and transmission power without needing knowledge of any specific network topology or route discovery process, and latency constraints are designed to meet the requirements of UAV applications. This method effectively improves the transmission reliability of UAV ad hoc networks in high-speed mobile environments and fast-fading channels, while reducing end-to-end latency and routing energy consumption.

[0058] The above embodiments are merely preferred embodiments of the present invention and should not be considered as limiting the scope of the present invention. All equivalent variations and improvements made within the scope of the present invention should still fall within the patent coverage of the present invention.

Claims

1. A low-energy, fast routing method for UAVs oriented towards task collaboration, characterized in that... Includes the following steps: Step 1: In the drone self-organizing network The number of data packets that need to be transmitted by the drone is: The time it takes to transmit one data packet is called one time slot, and the total number of time slots is [number missing]. Routing decisions based on CSMA / CA contention mechanism ,when Do not broadcast data packets, set up drones The number of supported transmission powers is ,when Transmission power Then the optional action ,in, Indicates the maximum transmission power; Step 2: Initialize the maximum routing efficiency weight Routing efficiency weight update rate The future benefits of the route learning process The weight of long-term returns shared by neighbors Delay risk level Risk threshold Latency risk learning rate State dimension This includes the maximum dimension of channel gain. , indivual Value matrix and Value matrix The weight of risk value in routing strategy selection The drone's own battery power Channel gain The set of received data packets , number of jumps Movement factor The number of times the monitored data packets were rebroadcast and initial performance signal-to-noise ratio and end-to-end delay ; Step 3: In the Time slots, drones Received drone Broadcast data packets Determine tuples Does it belong to the set of received data packets? ,in For the source drone ID, This is the data packet sequence number, which increments with each data packet sent by the source drone; if the tuple Belongs to the data packet set If the packet is not received, it should be discarded to avoid repeated broadcasting. Step 4: In the Time slots, drones Estimate the relationship with neighboring drones Channel gain Assess your remaining battery power. Signal-to-noise ratio of the received signal and data packet size Get the number of hops the current data packet has traversed from the IP packet header. Listen for the number of rebroadcasts of the previous data packet among the neighbors. Before obtaining destination feedback Average end-to-end latency of data packets Obtain the number of neighbors by interacting with neighbor beacons. Calculation of drones Compared to drones common neighbors set , Represent the neighbor set and calculate the mobility factor compared to the previous time slot. ; Step 5: Calculate the single-hop delay including contention and transmission delay. Channel gain Signal-to-noise ratio Single-hop delay and the number of rebroadcasts detected Encapsulated in the MAC packet header, the data packet is obtained through neighbor replay. Received signal signal-to-noise ratio Single-hop delay Statistics and Neighbors Channel gain Accumulate and update the number of rebroadcasts ; Step 6: Build the routing state ; Step 7: Configure routing status Input benefit table and risk table Obtain the long-term reward value of the state-action pair. and risk value Update strategy distribution : Step 8: Drone According to strategy distribution Perform routing decisions and power allocation, then process the tuples. Store in collection ; Step 9: After the target drone receives the data packet, it performs statistics. Average end-to-end latency of data packets The performance evaluation metrics are then sent to each drone via a reliable feedback channel; the drones Upon receiving routing performance metrics, it estimates its own transmission energy consumption. Calculate current benefits : in, This is a packet arrival indicator; if the data packet successfully reaches the destination drone, then... Otherwise, it is 0; and These represent the weighting factors for latency and energy consumption, respectively. Step 10: Drone Receive neighbor set Shared state-value function, updating the Q-table: The dynamic routing efficiency weight According to the movement factor Perform the calculation: Step 11: Drone Divide end-to-end delay into Risk Level , and the previous The latency of each data packet is compared, and the proportion of packets with latency less than the risk threshold is counted to obtain the latency tolerance probability. Based on the minimum latency tolerance... Calculate latency risk : in, This indicates an indicator function; it returns 1 if the condition in parentheses is true, and 0 otherwise; based on latency risk... Update table E: Step 12: Repeat steps 3 to 11 until the algorithm converges.

Images